RESEARCH ARTICLE
Utilization of smoking cessation medication benefits among medicaid fee-for-service enrollees 1999–2008 Jennifer Kahende1*, Ann Malarcher1, Lucinda England1, Lei Zhang1, Paul Mowery2, Xin Xu1, Varadan Sevilimedu2, Italia Rolle1 1 Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America, 2 Biostatistics Inc., Sarasota, Florida, United States of America
a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
* [email protected]
Abstract Objective
OPEN ACCESS Citation: Kahende J, Malarcher A, England L, Zhang L, Mowery P, Xu X, et al. (2017) Utilization of smoking cessation medication benefits among medicaid fee-for-service enrollees 1999–2008. PLoS ONE 12(2): e0170381. doi:10.1371/journal. pone.0170381 Editor: Takeru Abe, Yokohama City University, JAPAN Received: March 16, 2016 Accepted: January 4, 2017 Published: February 16, 2017 Copyright: This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication. Data Availability Statement: Data are available from the Centers for Disease Control and Prevention’s National Center for Health Statistics’ Research Data Center for researchers whose research proposal meets the criteria for access to confidential data. Information on how to access the data can be found at www.cdc.gov/rdc. Funding: The Centers for Disease Control and Prevention’s National Center for Chronic Disease Prevention and Health Promotion provided funding for this project (contract 4157-026). Through this
To assess state coverage and utilization of Medicaid smoking cessation medication benefits among fee-for-service enrollees who smoked cigarettes.
Methods We used the linked National Health Interview Survey (survey years 1995, 1997–2005) and the Medicaid Analytic eXtract files (1999–2008) to assess utilization of smoking cessation medication benefits among 5,982 cigarette smokers aged 18–64 years enrolled in Medicaid fee-for-service whose state Medicaid insurance covered at least one cessation medication. We excluded visits during pregnancy, and those covered by managed care or under dual enrollment (Medicaid and Medicare). Multivariate logistic regression was used to determine correlates of cessation medication benefit utilization among Medicaid fee-for-service enrollees, including measures of drug coverage (comprehensive cessation medication coverage, number of medications in state benefit, varenicline coverage), individual-level demographics at NHIS interview, age at Medicaid enrollment, and state-level cigarette excise taxes, statewide smoke-free laws, and per-capita tobacco control funding.
Results In 1999, the percent of smokers with 1 medication claims was 5.7% in the 30 states that covered at least one Food and Drug Administration (FDA)-approved cessation medication; this increased to 9.9% in 2008 in the 44 states that covered at least one FDA-approved medication (p1 day in the past year compared with 55.4% in the general population [3]; and 5.9% were recent successful quitters compared with 7.4% in the general population. Medicaid-enrolled smokers are also as likely as those in the general population to use cessation treatments [3]. Among current cigarette smokers enrolled in Medicaid who tried to quit in the past year and former smokers who successfully quit in the past 2 years, 32.2% had used a cessation medication and 8.0% used counseling, compared with 29.0% and 6.8%, respectively, of smokers in the general population [3]. Insurance coverage of evidence-based smoking cessation treatments is associated with increased quit attempts, use of cessation treatments, and successful smoking cessation [4]. In 2011, Healthy People 2020 was established to provide science-based, 10-year national objectives for improving the health of all Americans [5]. One of the Healthy People 2020 objectives (TU8) is to increase the number of states whose Medicaid insurance covers all evidence-based smoking cessation treatments including individual, group, and telephone counseling and the seven Food and Drug Administration (FDA)-approved cessation medications [5]. Although progress toward meeting this objective has occurred, many states lack comprehensive cessation coverage. For example, the number of states that covered any counseling increased from 7 in 1999 to 49 in 2014 (however, five of the 49 states restricted counseling to only pregnant women) [6,7,8,9]. Additionally, the number that covered all FDA-approved cessation medications increased from 13 in 1999 to 30 in 2015 [6,7,8,9]. Although the 2010 Affordable Care Act prohibits state Medicaid programs from excluding coverage for FDA-approved smoking cessation medications, including over-the-counter medications, effective January 1, 2014 [10], as of June 30, 2015, based on data from the American Lung Association, 30 state Medicaid programs covered all seven FDA-approved cessation medications [9]. Large state variations have been observed in cessation medication utilization among Medicaid enrollees. For example, a recent study by Ku et al. using Medicaid’s 2013 aggregate state prescription drug utilization rebate files, estimated that utilization of tobacco cessation medication among Medicaid adult smokers varied from 1% in Texas to 27% in Minnesota [11]. To
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
2 / 15
Smoking cessation medication utilization among medicaid enrollees
date, few studies have examined the relationship between smoking cessation treatment coverage and cessation outcomes within the Medicaid population. Greene et al. [12] found that enrollees residing in states with the most extensive coverage (counseling and pharmacotherapy with no copayment) had higher rates of successfully quitting in the past 12 months as compared to enrollees residing in states with no coverage and those residing in states in which copayments were required for both counseling and pharmacotherapy. Moreover, Li and Dresler [13] found that the number of tobacco cessation drug claims increased sharply after coverage for nicotine patch and gum started in Arkansas in October 2004, peaked in December 2004, and then declined rapidly; a similar pattern was observed with the addition of varenicline coverage. Their study concluded that Medicaid coverage alone may have a limited, sustained effect on increasing utilization of covered cessation treatments [13]. In addition, Liu [14] observed that smokers residing in states with smoking cessation treatment coverage and no copayment requirements were more likely to have made a quit attempt in the past 12 months and to report intention to quit in the next 6 months than smokers residing in states with no tobacco-dependency treatment coverage. Finally, when Massachusetts widely promoted their new comprehensive cessation benefit coverage to Medicaid enrollees and health care providers, 37% of all enrollees who smoked used the cessation benefit, and cigarette smoking prevalence decreased significantly from 38% in the pre-benefit period to 28% in the post-benefit period [15]. There are currently no studies examining the overall prevalence of utilization of smoking cessation medication benefits among the Medicaid population using individual claims data and how benefit utilization varies by the number or type of medications covered or other individual and state-level characteristics. To fill this gap, we assessed trends in the utilization of smoking cessation medication benefits during 1999–2008 among Medicaid fee-for-service enrollees whose state Medicaid insurance covered at least one smoking cessation medication. We also examined the relationship between cessation medication benefit utilization and coverage of all FDA-approved medications in a given year, the number of medications covered, coverage of varenicline (Chantix) after 2006, individual-level demographic characteristics of Medicaid smokers, and state-level characteristics (cigarette excise tax, statewide smoke-free laws, and per-capita tobacco control program funding).
Methods Data The Centers for Disease Control and Prevention’s National Center for Health Promotion and Disease Prevention’s and its National Center for Health Statistics’ (NCHS) IRB authorities reviewed and approved this study. The study was performed at the NCHS’s Research Data Center and no individual level information was released to the study authors. We used data from the 1995 and 1997–2005 National Health Interview Surveys (NHIS) linked to the 1999– 2008 Centers for Medicare & Medicaid Services (CMS) Medicaid Analytic eXtract files [16]; no tobacco use questions were included on the NHIS 1996 survey, so subjects interviewed in this year were not included in our analysis. The NHIS utilizes a complex probability sample design to select a nationally representative sample of the civilian, non-institutionalized population of the United States. The household-based survey consists of a core questionnaire with demographic and health-related questions, as well as supplemental questionnaires on a variety of health-related topics. The Medicaid MAX files include Medicaid enrollment and claims data for NHIS respondents who were successfully matched by the National Center for Health Statistics with Medicaid administrative records. Details of the linkage methodologies and the linked data files are published elsewhere [16].
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
3 / 15
Smoking cessation medication utilization among medicaid enrollees
Study population The study population included all NHIS respondents 18–64 years of age during a Medicaid enrollment year who were either current or former cigarette smokers at the time of the NHIS interview and were residents of states whose Medicaid program covered a least one smoking cessation medication at any time from 1999 through 2008. We then limited the population to those who were enrolled in fee-for-service Medicaid for at least one month during the study period, according to the Medicaid claims data. Details on how we obtained our study population are included in Figure A in S1 Appendix. For our analysis, we examined only periods of enrollment in Medicaid fee-for-service programs, and excluded the data for the entire enrollment year if a person was enrolled in either Medicaid managed care or had dual enrollment in Medicaid and Medicare during any month in a 12-month calendar year. These exclusions were made because for dual enrollees, Medicare is the first payer for services that are covered by both Medicare and Medicaid; MAX only captures these services if additional Medicaid payments are made on behalf of the enrollee for either Medicare cost-sharing or for shared services. In addition, encounter records for those enrollees in Medicaid managed care plans are not provided by all states; in some states, only a portion of managed care recipients have encounter data recorded. Because cessation medication recommendations differ for pregnant women [4] and because charges acquired during pregnancy are often bundled, we also excluded monthly periods of enrollment during pregnancy.
Cigarette smoking status Self-reported cigarette smoking information from the NHIS was used to identify respondents as current smokers or former smokers at the time of their interview. Current smokers were defined as those who answered “yes” when asked “Have you smoked at least 100 cigarettes in your entire life?”, and who also answered “every day” or “some days” when asked “Do you now smoke cigarettes every day, some days or not at all?” Former smokers were defined as those who ever smoked 100 cigarettes, but who answered “not at all” to the aforementioned question. We constructed the person’s past smoking history from NHIS using age of initiation and time since quitting to determine whether they were current smokers at the time of their Medicaid claim. For example, if a person reported that they were a former smoker at time of NHIS interview in 2005, had initiated smoking in 1980 and quit in 2004 then we defined them as a current smoker for any Medicaid claims in the years 1999 (the year the Medicaid claims began) through 2004. For those who were current smokers at the time of their Medicaid claim according to their smoking history, we included Medicaid claims data from years prior to the NHIS interview. For Medicaid claims that occurred after the person’s NHIS interview, we assumed smoking status at the time of the NHIS interview remained constant for the remainder of the study period; only claims for those categorized as ‘current smokers’ at time of Medicaid claim were included.
Medicaid claims for cigarette smoking cessation medications The primary outcome measure for this study was whether a current smoker had one or more smoking cessation medication claims in a Medicaid enrollment year. Because enrollees could appear in multiple years, drug claims were assessed for each enrollment year from 1999 to 2008. The percent of Medicaid enrollees with 1 cessation medication claim in a given year was calculated. We included all FDA- approved smoking cessation medications (nicotine patch, gum, spray, inhaler, bupropion, lozenges, and varenicline) starting in the year they were approved. Since bupropion is also commonly used for the treatment of depression and other
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
4 / 15
Smoking cessation medication utilization among medicaid enrollees
conditions, we also performed sensitivity analyses by excluding bupropion claims from all analysis. Details on drug coding are available in the Text A in S1 Appendix.
Individual measures Self-reported gender (male, female), race/ethnicity (Non-Hispanic (NH) White, NH Black, Hispanic, and NH “other”), and educational attainment (< High School, High School/GED, and > High School) of Medicaid enrollees were obtained from the NHIS. Age (18–24, 25–34, 35–44, 45–64 years) was age at Medicaid enrollment year.
State level measures Three measures of cessation medication coverage were constructed for each year of the study period: (1) comprehensive drug coverage (i.e., covered all FDA-approved medications; yes/ no); (2) number of drugs covered; and (3) varenicline coverage (yes/no). Comprehensive drug coverage was defined as coverage of 5 drugs during 1999–2001, 6 drugs during 2002–2005 (nicotine lozenges were approved in 2002), and 7 drugs during 2006–2008 (varenicline was approved in 2006). Information on state Medicaid drug coverage during 1999–2007 came from published Morbidity Mortality Weekly Reports [6,7,8,17,18,19]; data for 2008 were obtained from the American Lung Association [20]. Three state-level variables thought to potentially influence the likelihood of using smoking cessation medications were included in the analysis: 1) state excise taxes on cigarettes (per pack of cigarettes on November 1 of each year [21], adjusted for inflation using 1999 index year); 2) per-capita funding for comprehensive statewide tobacco control programs; and 3) presence or absence of a comprehensive statewide smoke-free law. State funding for tobacco control programs was obtained from the Bridging the Gap/ImpacTeen Project, University of Illinois at Chicago Health Policy Center, [22] where estimates were calculated from the following sources: (1) state appropriations and allocations to tobacco control programs generated from state tobacco excise tax revenues, state settlement payments, and/or state general funds; (2) federal funds received by states and communities for tobacco control and prevention; (3) tobacco control funds from the Robert Wood Johnson Foundation; and (4) tobacco control funds from the American Legacy Foundation. Funding was calculated per capita and was adjusted for inflation using 1999 as the index year [23]. Statewide comprehensive smoke-free laws were determined using the State Tobacco Activities Tracking and Evaluation (STATE) System [23]; states were categorized dichotomously as having a comprehensive smoke-free law covering private workplaces, restaurants and bars versus any other less restrictive policy or no policy. All three state level variables were calculated for each Medicaid enrollment year during 1999–2008.
Statistical analysis Three analyses were carried out: 1) descriptive statistics showing the percentage of enrollees who had 1 claim for a smoking cessation medication in a year, by year and demographic characteristics; 2) time trend tests of the percentages during 1999–2008; and 3) logistic regression to test for independent associations between a Medicaid enrollee’s cessation medication utilization and state Medicaid cessation medication coverage, individual demographic characteristics (age, gender, race/ethnicity, and education), state excise tax, per-capita state tobacco control program funding, and comprehensive state smoke-free law status. The percentage of enrollees with 1 medication claim was calculated by summing the indicator variable (whether a person had a cessation medication claim in a given year, 1 = yes, 0 = no) for each enrollment year and dividing that figure by the total number of enrollees who were enrolled for at least one
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
5 / 15
Smoking cessation medication utilization among medicaid enrollees
month during that year. Time trends for the annual percentage of enrollees having a claim were carried out using least squares regression and assuming a linear relationship over time. The outcome measure for the logistic models was whether a smoker had one or more smoking cessation medication claims in a Medicaid enrollment year. The three different cessation medication coverage measures were evaluated individually in separate models. Survey weights were used to account for the probability of selection for each NHIS respondent and a nonresponse adjustment was used to account for demographic differences in linkage eligibility of the linked-MAX files. A post-stratification adjustment was then applied to make the analysis sample demographic distributions match the total MAX file population, by enrollment year. Data were also weighted by the number of enrollment months to control for differences in length of enrollment and eligibility to make a claim. Stratification and cluster variables were included in the analyses to take into account the NHIS multi-stage survey design as well as the fact that respondents could appear in multiple years of the MAX files. SAS1 version 9.3 was used for all analyses.
Results The number of states that covered at least one drug increased from 30 in 1999 to 45 in 2008, and the number of states that offered comprehensive drug coverage increased from 13 in 1999 (5 drugs) to 24 in 2008 (7 drugs) (Table 1). During 2000–2008, at least 50% of states that covered one or more smoking cessation drugs had comprehensive coverage. The Centers for Medicare and Medicaid Services estimated that those enrolled in fee-for-service represented 59%-61% of the Medicaid population during the midpoint of this study [24]. Among our Medicaid fee-for-service study population, only around one-half were enrolled in Medicaid fee-for-service coverage for the entire year (range: 48.7% in 1999 to 53.0% in 2004; data not shown) and on average they were enrolled for 8.8 months per year (range 8.6 in 1999 to 9.0 in 2003; data not shown). Table 2 shows demographic characteristics of study participants by whether they had a cessation medication claim. Compared with enrollees without a claim, enrollees with a cessation medication claim were more likely to be in older age groups at the Table 1. Number of states whose Medicaid program covered FDA-approved smoking cessation medications,1 by number of medications covered and year, 1999–2008.2 Number of Medications1
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008 0
1
2
2
3
3
4
4
3
0
0
2
2
2
2
2
1
1
1
2
1
0
3
12
11
10
6
8
7
6
3
3
4 4
4
1
1
1
1
1
1
6
2
2
5
13
16
18
21
20
20
23
5
4
6
6
NA
NA
NA
0
0
0
0
8
5
7
7
NA
NA
NA
NA
NA
NA
NA
19
26
24
30
32
34
33
34
33
39
39
41
45
Total Abbreviations: NA = Not Applicable 1
FDA approved nicotine patch, gum, spray in 1996; inhaler and bupropion in 1997; nicotine lozenges in 2002 and varenicline in 2006.
2
Drug coverage data for 1999–2007 came from published Morbidity Mortality Weekly Reports (MMWR 1998–2000, 1994–2001, 1994–2002, 2005, 2006, 2007) and 2008 data were obtained from the American Lung Association (ALA 2008)[6–8,17–20]. Coverage data from 2002 and 2005 was used to fill in
missing information for 2003 and 2004: if there was no change in coverage between 2002 and 2005, we filled in the same information for 2003 and 2004; if 2002 coverage data was different from 2005, we used 2002 data to fill in the missing information for 2003 and 2004, assuming that coverage remained the same until 2005. We excluded the drug coverage if it was for pregnant women-only or for Medicaid Managed Care-only. doi:10.1371/journal.pone.0170381.t001
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
6 / 15
Smoking cessation medication utilization among medicaid enrollees
Table 2. Demographic characteristics1 of select Medicaid fee-for-service enrollees2 who were current cigarette smokers,3 by whether they had a smoking cessation medication claim4: NHIS (1995, 1997–2005) linked to Medicaid Max files (1999–2008). Enrollees with 1 cessation medication Enrollees with no cessation medication claim claim Characteristic and Data Source
P-value
Age (years) (Max)
P
Utilization of smoking cessation medication benefits among medicaid fee-for-service enrollees 1999–2008 Jennifer Kahende1*, Ann Malarcher1, Lucinda England1, Lei Zhang1, Paul Mowery2, Xin Xu1, Varadan Sevilimedu2, Italia Rolle1 1 Centers for Disease Control and Prevention, Atlanta, Georgia, United States of America, 2 Biostatistics Inc., Sarasota, Florida, United States of America
a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
* [email protected]
Abstract Objective
OPEN ACCESS Citation: Kahende J, Malarcher A, England L, Zhang L, Mowery P, Xu X, et al. (2017) Utilization of smoking cessation medication benefits among medicaid fee-for-service enrollees 1999–2008. PLoS ONE 12(2): e0170381. doi:10.1371/journal. pone.0170381 Editor: Takeru Abe, Yokohama City University, JAPAN Received: March 16, 2016 Accepted: January 4, 2017 Published: February 16, 2017 Copyright: This is an open access article, free of all copyright, and may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose. The work is made available under the Creative Commons CC0 public domain dedication. Data Availability Statement: Data are available from the Centers for Disease Control and Prevention’s National Center for Health Statistics’ Research Data Center for researchers whose research proposal meets the criteria for access to confidential data. Information on how to access the data can be found at www.cdc.gov/rdc. Funding: The Centers for Disease Control and Prevention’s National Center for Chronic Disease Prevention and Health Promotion provided funding for this project (contract 4157-026). Through this
To assess state coverage and utilization of Medicaid smoking cessation medication benefits among fee-for-service enrollees who smoked cigarettes.
Methods We used the linked National Health Interview Survey (survey years 1995, 1997–2005) and the Medicaid Analytic eXtract files (1999–2008) to assess utilization of smoking cessation medication benefits among 5,982 cigarette smokers aged 18–64 years enrolled in Medicaid fee-for-service whose state Medicaid insurance covered at least one cessation medication. We excluded visits during pregnancy, and those covered by managed care or under dual enrollment (Medicaid and Medicare). Multivariate logistic regression was used to determine correlates of cessation medication benefit utilization among Medicaid fee-for-service enrollees, including measures of drug coverage (comprehensive cessation medication coverage, number of medications in state benefit, varenicline coverage), individual-level demographics at NHIS interview, age at Medicaid enrollment, and state-level cigarette excise taxes, statewide smoke-free laws, and per-capita tobacco control funding.
Results In 1999, the percent of smokers with 1 medication claims was 5.7% in the 30 states that covered at least one Food and Drug Administration (FDA)-approved cessation medication; this increased to 9.9% in 2008 in the 44 states that covered at least one FDA-approved medication (p1 day in the past year compared with 55.4% in the general population [3]; and 5.9% were recent successful quitters compared with 7.4% in the general population. Medicaid-enrolled smokers are also as likely as those in the general population to use cessation treatments [3]. Among current cigarette smokers enrolled in Medicaid who tried to quit in the past year and former smokers who successfully quit in the past 2 years, 32.2% had used a cessation medication and 8.0% used counseling, compared with 29.0% and 6.8%, respectively, of smokers in the general population [3]. Insurance coverage of evidence-based smoking cessation treatments is associated with increased quit attempts, use of cessation treatments, and successful smoking cessation [4]. In 2011, Healthy People 2020 was established to provide science-based, 10-year national objectives for improving the health of all Americans [5]. One of the Healthy People 2020 objectives (TU8) is to increase the number of states whose Medicaid insurance covers all evidence-based smoking cessation treatments including individual, group, and telephone counseling and the seven Food and Drug Administration (FDA)-approved cessation medications [5]. Although progress toward meeting this objective has occurred, many states lack comprehensive cessation coverage. For example, the number of states that covered any counseling increased from 7 in 1999 to 49 in 2014 (however, five of the 49 states restricted counseling to only pregnant women) [6,7,8,9]. Additionally, the number that covered all FDA-approved cessation medications increased from 13 in 1999 to 30 in 2015 [6,7,8,9]. Although the 2010 Affordable Care Act prohibits state Medicaid programs from excluding coverage for FDA-approved smoking cessation medications, including over-the-counter medications, effective January 1, 2014 [10], as of June 30, 2015, based on data from the American Lung Association, 30 state Medicaid programs covered all seven FDA-approved cessation medications [9]. Large state variations have been observed in cessation medication utilization among Medicaid enrollees. For example, a recent study by Ku et al. using Medicaid’s 2013 aggregate state prescription drug utilization rebate files, estimated that utilization of tobacco cessation medication among Medicaid adult smokers varied from 1% in Texas to 27% in Minnesota [11]. To
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
2 / 15
Smoking cessation medication utilization among medicaid enrollees
date, few studies have examined the relationship between smoking cessation treatment coverage and cessation outcomes within the Medicaid population. Greene et al. [12] found that enrollees residing in states with the most extensive coverage (counseling and pharmacotherapy with no copayment) had higher rates of successfully quitting in the past 12 months as compared to enrollees residing in states with no coverage and those residing in states in which copayments were required for both counseling and pharmacotherapy. Moreover, Li and Dresler [13] found that the number of tobacco cessation drug claims increased sharply after coverage for nicotine patch and gum started in Arkansas in October 2004, peaked in December 2004, and then declined rapidly; a similar pattern was observed with the addition of varenicline coverage. Their study concluded that Medicaid coverage alone may have a limited, sustained effect on increasing utilization of covered cessation treatments [13]. In addition, Liu [14] observed that smokers residing in states with smoking cessation treatment coverage and no copayment requirements were more likely to have made a quit attempt in the past 12 months and to report intention to quit in the next 6 months than smokers residing in states with no tobacco-dependency treatment coverage. Finally, when Massachusetts widely promoted their new comprehensive cessation benefit coverage to Medicaid enrollees and health care providers, 37% of all enrollees who smoked used the cessation benefit, and cigarette smoking prevalence decreased significantly from 38% in the pre-benefit period to 28% in the post-benefit period [15]. There are currently no studies examining the overall prevalence of utilization of smoking cessation medication benefits among the Medicaid population using individual claims data and how benefit utilization varies by the number or type of medications covered or other individual and state-level characteristics. To fill this gap, we assessed trends in the utilization of smoking cessation medication benefits during 1999–2008 among Medicaid fee-for-service enrollees whose state Medicaid insurance covered at least one smoking cessation medication. We also examined the relationship between cessation medication benefit utilization and coverage of all FDA-approved medications in a given year, the number of medications covered, coverage of varenicline (Chantix) after 2006, individual-level demographic characteristics of Medicaid smokers, and state-level characteristics (cigarette excise tax, statewide smoke-free laws, and per-capita tobacco control program funding).
Methods Data The Centers for Disease Control and Prevention’s National Center for Health Promotion and Disease Prevention’s and its National Center for Health Statistics’ (NCHS) IRB authorities reviewed and approved this study. The study was performed at the NCHS’s Research Data Center and no individual level information was released to the study authors. We used data from the 1995 and 1997–2005 National Health Interview Surveys (NHIS) linked to the 1999– 2008 Centers for Medicare & Medicaid Services (CMS) Medicaid Analytic eXtract files [16]; no tobacco use questions were included on the NHIS 1996 survey, so subjects interviewed in this year were not included in our analysis. The NHIS utilizes a complex probability sample design to select a nationally representative sample of the civilian, non-institutionalized population of the United States. The household-based survey consists of a core questionnaire with demographic and health-related questions, as well as supplemental questionnaires on a variety of health-related topics. The Medicaid MAX files include Medicaid enrollment and claims data for NHIS respondents who were successfully matched by the National Center for Health Statistics with Medicaid administrative records. Details of the linkage methodologies and the linked data files are published elsewhere [16].
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
3 / 15
Smoking cessation medication utilization among medicaid enrollees
Study population The study population included all NHIS respondents 18–64 years of age during a Medicaid enrollment year who were either current or former cigarette smokers at the time of the NHIS interview and were residents of states whose Medicaid program covered a least one smoking cessation medication at any time from 1999 through 2008. We then limited the population to those who were enrolled in fee-for-service Medicaid for at least one month during the study period, according to the Medicaid claims data. Details on how we obtained our study population are included in Figure A in S1 Appendix. For our analysis, we examined only periods of enrollment in Medicaid fee-for-service programs, and excluded the data for the entire enrollment year if a person was enrolled in either Medicaid managed care or had dual enrollment in Medicaid and Medicare during any month in a 12-month calendar year. These exclusions were made because for dual enrollees, Medicare is the first payer for services that are covered by both Medicare and Medicaid; MAX only captures these services if additional Medicaid payments are made on behalf of the enrollee for either Medicare cost-sharing or for shared services. In addition, encounter records for those enrollees in Medicaid managed care plans are not provided by all states; in some states, only a portion of managed care recipients have encounter data recorded. Because cessation medication recommendations differ for pregnant women [4] and because charges acquired during pregnancy are often bundled, we also excluded monthly periods of enrollment during pregnancy.
Cigarette smoking status Self-reported cigarette smoking information from the NHIS was used to identify respondents as current smokers or former smokers at the time of their interview. Current smokers were defined as those who answered “yes” when asked “Have you smoked at least 100 cigarettes in your entire life?”, and who also answered “every day” or “some days” when asked “Do you now smoke cigarettes every day, some days or not at all?” Former smokers were defined as those who ever smoked 100 cigarettes, but who answered “not at all” to the aforementioned question. We constructed the person’s past smoking history from NHIS using age of initiation and time since quitting to determine whether they were current smokers at the time of their Medicaid claim. For example, if a person reported that they were a former smoker at time of NHIS interview in 2005, had initiated smoking in 1980 and quit in 2004 then we defined them as a current smoker for any Medicaid claims in the years 1999 (the year the Medicaid claims began) through 2004. For those who were current smokers at the time of their Medicaid claim according to their smoking history, we included Medicaid claims data from years prior to the NHIS interview. For Medicaid claims that occurred after the person’s NHIS interview, we assumed smoking status at the time of the NHIS interview remained constant for the remainder of the study period; only claims for those categorized as ‘current smokers’ at time of Medicaid claim were included.
Medicaid claims for cigarette smoking cessation medications The primary outcome measure for this study was whether a current smoker had one or more smoking cessation medication claims in a Medicaid enrollment year. Because enrollees could appear in multiple years, drug claims were assessed for each enrollment year from 1999 to 2008. The percent of Medicaid enrollees with 1 cessation medication claim in a given year was calculated. We included all FDA- approved smoking cessation medications (nicotine patch, gum, spray, inhaler, bupropion, lozenges, and varenicline) starting in the year they were approved. Since bupropion is also commonly used for the treatment of depression and other
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
4 / 15
Smoking cessation medication utilization among medicaid enrollees
conditions, we also performed sensitivity analyses by excluding bupropion claims from all analysis. Details on drug coding are available in the Text A in S1 Appendix.
Individual measures Self-reported gender (male, female), race/ethnicity (Non-Hispanic (NH) White, NH Black, Hispanic, and NH “other”), and educational attainment (< High School, High School/GED, and > High School) of Medicaid enrollees were obtained from the NHIS. Age (18–24, 25–34, 35–44, 45–64 years) was age at Medicaid enrollment year.
State level measures Three measures of cessation medication coverage were constructed for each year of the study period: (1) comprehensive drug coverage (i.e., covered all FDA-approved medications; yes/ no); (2) number of drugs covered; and (3) varenicline coverage (yes/no). Comprehensive drug coverage was defined as coverage of 5 drugs during 1999–2001, 6 drugs during 2002–2005 (nicotine lozenges were approved in 2002), and 7 drugs during 2006–2008 (varenicline was approved in 2006). Information on state Medicaid drug coverage during 1999–2007 came from published Morbidity Mortality Weekly Reports [6,7,8,17,18,19]; data for 2008 were obtained from the American Lung Association [20]. Three state-level variables thought to potentially influence the likelihood of using smoking cessation medications were included in the analysis: 1) state excise taxes on cigarettes (per pack of cigarettes on November 1 of each year [21], adjusted for inflation using 1999 index year); 2) per-capita funding for comprehensive statewide tobacco control programs; and 3) presence or absence of a comprehensive statewide smoke-free law. State funding for tobacco control programs was obtained from the Bridging the Gap/ImpacTeen Project, University of Illinois at Chicago Health Policy Center, [22] where estimates were calculated from the following sources: (1) state appropriations and allocations to tobacco control programs generated from state tobacco excise tax revenues, state settlement payments, and/or state general funds; (2) federal funds received by states and communities for tobacco control and prevention; (3) tobacco control funds from the Robert Wood Johnson Foundation; and (4) tobacco control funds from the American Legacy Foundation. Funding was calculated per capita and was adjusted for inflation using 1999 as the index year [23]. Statewide comprehensive smoke-free laws were determined using the State Tobacco Activities Tracking and Evaluation (STATE) System [23]; states were categorized dichotomously as having a comprehensive smoke-free law covering private workplaces, restaurants and bars versus any other less restrictive policy or no policy. All three state level variables were calculated for each Medicaid enrollment year during 1999–2008.
Statistical analysis Three analyses were carried out: 1) descriptive statistics showing the percentage of enrollees who had 1 claim for a smoking cessation medication in a year, by year and demographic characteristics; 2) time trend tests of the percentages during 1999–2008; and 3) logistic regression to test for independent associations between a Medicaid enrollee’s cessation medication utilization and state Medicaid cessation medication coverage, individual demographic characteristics (age, gender, race/ethnicity, and education), state excise tax, per-capita state tobacco control program funding, and comprehensive state smoke-free law status. The percentage of enrollees with 1 medication claim was calculated by summing the indicator variable (whether a person had a cessation medication claim in a given year, 1 = yes, 0 = no) for each enrollment year and dividing that figure by the total number of enrollees who were enrolled for at least one
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
5 / 15
Smoking cessation medication utilization among medicaid enrollees
month during that year. Time trends for the annual percentage of enrollees having a claim were carried out using least squares regression and assuming a linear relationship over time. The outcome measure for the logistic models was whether a smoker had one or more smoking cessation medication claims in a Medicaid enrollment year. The three different cessation medication coverage measures were evaluated individually in separate models. Survey weights were used to account for the probability of selection for each NHIS respondent and a nonresponse adjustment was used to account for demographic differences in linkage eligibility of the linked-MAX files. A post-stratification adjustment was then applied to make the analysis sample demographic distributions match the total MAX file population, by enrollment year. Data were also weighted by the number of enrollment months to control for differences in length of enrollment and eligibility to make a claim. Stratification and cluster variables were included in the analyses to take into account the NHIS multi-stage survey design as well as the fact that respondents could appear in multiple years of the MAX files. SAS1 version 9.3 was used for all analyses.
Results The number of states that covered at least one drug increased from 30 in 1999 to 45 in 2008, and the number of states that offered comprehensive drug coverage increased from 13 in 1999 (5 drugs) to 24 in 2008 (7 drugs) (Table 1). During 2000–2008, at least 50% of states that covered one or more smoking cessation drugs had comprehensive coverage. The Centers for Medicare and Medicaid Services estimated that those enrolled in fee-for-service represented 59%-61% of the Medicaid population during the midpoint of this study [24]. Among our Medicaid fee-for-service study population, only around one-half were enrolled in Medicaid fee-for-service coverage for the entire year (range: 48.7% in 1999 to 53.0% in 2004; data not shown) and on average they were enrolled for 8.8 months per year (range 8.6 in 1999 to 9.0 in 2003; data not shown). Table 2 shows demographic characteristics of study participants by whether they had a cessation medication claim. Compared with enrollees without a claim, enrollees with a cessation medication claim were more likely to be in older age groups at the Table 1. Number of states whose Medicaid program covered FDA-approved smoking cessation medications,1 by number of medications covered and year, 1999–2008.2 Number of Medications1
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008 0
1
2
2
3
3
4
4
3
0
0
2
2
2
2
2
1
1
1
2
1
0
3
12
11
10
6
8
7
6
3
3
4 4
4
1
1
1
1
1
1
6
2
2
5
13
16
18
21
20
20
23
5
4
6
6
NA
NA
NA
0
0
0
0
8
5
7
7
NA
NA
NA
NA
NA
NA
NA
19
26
24
30
32
34
33
34
33
39
39
41
45
Total Abbreviations: NA = Not Applicable 1
FDA approved nicotine patch, gum, spray in 1996; inhaler and bupropion in 1997; nicotine lozenges in 2002 and varenicline in 2006.
2
Drug coverage data for 1999–2007 came from published Morbidity Mortality Weekly Reports (MMWR 1998–2000, 1994–2001, 1994–2002, 2005, 2006, 2007) and 2008 data were obtained from the American Lung Association (ALA 2008)[6–8,17–20]. Coverage data from 2002 and 2005 was used to fill in
missing information for 2003 and 2004: if there was no change in coverage between 2002 and 2005, we filled in the same information for 2003 and 2004; if 2002 coverage data was different from 2005, we used 2002 data to fill in the missing information for 2003 and 2004, assuming that coverage remained the same until 2005. We excluded the drug coverage if it was for pregnant women-only or for Medicaid Managed Care-only. doi:10.1371/journal.pone.0170381.t001
PLOS ONE | DOI:10.1371/journal.pone.0170381 February 16, 2017
6 / 15
Smoking cessation medication utilization among medicaid enrollees
Table 2. Demographic characteristics1 of select Medicaid fee-for-service enrollees2 who were current cigarette smokers,3 by whether they had a smoking cessation medication claim4: NHIS (1995, 1997–2005) linked to Medicaid Max files (1999–2008). Enrollees with 1 cessation medication Enrollees with no cessation medication claim claim Characteristic and Data Source
P-value
Age (years) (Max)
P
